Conventionally, information indicative of an object moving in an image photographed by an image pick-up device such as a camera (which will be hereinafter referred to as motion information) has been utilized in various image processings. As an example, a cyclic type noise reduction processing of reducing a noise of an image is taken. The cyclic type noise reduction processing is a variable parameter low-pass filter in a time direction of an animation and can be expressed in the following transfer function by using a frame memory in one frame.H(zF(0))=(1−K)/(1−KzF(−1))wherein zF(−1) represents a unit delay operator for one frame.
The cyclic type noise reduction processing is the low-pass filter in the time direction as described above. Therefore, a moving portion is blurred on a time basis. A parameter K is varied within a range of 1>K>0 in the moving portion of an image in such a manner that the blur is included within a range of a visual afterimage and can be prevented from being a visual disturbance. In this case, K→1 is set in a stationary portion (the blur is not generated on principle) and K→0 is set in the moving portion (the blur is not generated and a noise reducing effect is deteriorated). In addition, the motion information is utilized for various uses, for example, “a three-dimensional luminance color separation”, “an interlace progressive conversion”, “a moving object detection” and “coding”.
In some conventional image processing apparatuses, a luminance color signal separating circuit and a noise reducing circuit are integrated with each other to share a memory (for example, see Patent Document 1). FIG. 5 is a block diagram showing schematic structures of the luminance color signal separating circuit and the noise reducing circuit which have been disclosed in the Patent Document 1. In FIG. 5, a composite video input signal supplied to an input terminal 1 is sent to adders 3 and 4 through a (1-K)-time weighting circuit 2. Moreover, a signal sent from the weighting circuit 2 is stored in a memory 6 through a selector 5 and a signal sent one frame before is supplied to the adder 3 through an inverting circuit 7, for example, and is directly supplied to the adder 4. Signals sent from the adders 3 and 4 are fetched from output terminals 8 and 9, respectively. Moreover, the signals sent from the adders 3 and 4 are added in an adder 12 through K-time weighting circuits 10 and 11 respectively and are stored in the memory 6 through the selector 5, and the switching of the selector 5 is controlled by a motion detecting signal supplied to a terminal 13. Then, a signal sent from the weighting circuit 2 is selected for a pixel subjected to a motion detection of the composite video input signal, for example.
Referring to a pixel from which the motion is not detected, an irregular noise component is reduced, and furthermore, a processing of separating a luminance signal and a color signal from the composite video signal is carried out. By increasing a value of the weighting K, it is possible to enhance the noise reducing effect. More specifically, a pixel in which a noise component is reduced is restricted to a pixel from which a motion is not detected. By increasing the value of K, therefore, it is possible to enhance the noise reducing effect. In the image processing apparatus, the memory is shared so that a cost can be reduced (Patent Document 1).
Patent Document 1: JP-A-2002-330447